Table 3.

Performance of the matching algorithm using the 4Mycotoxins training set (1,338 sequences) and the 97AerobiotaSamples testing set by signature length λ

aodp
λ	μ98	$\overline{\Theta }/\overline{\Psi }$	$\overline{\Psi }$	$\overline{\Omega }$	t
16	1352	0.93	0.317	17.41	17039
24	1353	0.94	0.311	13.27	9720
32	1342	0.95	0.299	11.83	6362
40	1325	0.94	0.298	11.06	3031
USEARCH					32560
BLAST					74335

μ98: number of matching query sequences with similarity α≥1−2ε=0.98, t: running time in seconds (system description in “Comparisons with other algorithms” section). Average values (algorithm 1) are reported for: size of the matching kernel $\overline{\Psi }$, number of sequences in all matching clusters $\overline{\Omega }$. Ratio $\overline{\Theta }/\overline{\Psi }$: average size of the result set to the average size of the matching kernel. Running times are also reported for USEARCH and BLAST